What Is a Network Fabric?
Modern networking is complicated. It’s challenging enough that you have Wi-Fi routers that can handle multiple hundreds of devices, the Internet of Things putting even more devices into a group, and the sheer magnitude of data centers and high-speed networking.
You also have virtualization, which allows you to take a complicated network and add entire dimensions to the design.
It’s a lot, and keeping up with everything as a network designer or manager requires a consolidated outlook.
That brings us to the network fabric. If you understand network fabrics, you have more options, and it’s a lot easier to keep track of everything.
What Is a Network Fabric?
A network fabric is not so much a thing as it is a way to think about networking. The fabric consists of the entire mesh of connections between every network device in a group or network. In a simple wired network, it’s a map of all of the wires connecting the various devices.
In modern networks, the fabric is the combined array of all of the wired, wireless, and virtual connections in the network.
In this way, a network fabric is similar to a network topology. The topology is a visual representation of every connection and node in the network. For a modern network with high levels of interconnectivity, that map will start to look a lot like a fabric weave — hence the name network fabric.
Perhaps the best way to think about a fabric is that it’s a network design philosophy that tries to maximize the number of stable connections between devices in a network. By having high levels of interconnectivity like this, you get more possible communication pathways. This allows your routing devices to maximize the speed of every transmission. It also creates a more robust network. If some pathways become unavailable, the routers can simply utilize other pathways to compensate.
What Does It Really Do for You?
There are a few ways to think about network fabrics, and each provides value as a network designer or administrator.
The first is how you might learn about fabrics in a classroom. The network fabric is the conceptual understanding of how all of your devices connect across the network. When you have a strong conceptualization, it’s easy to see how and where you might want to make adjustments within the fabric to ensure you have the right levels of connectivity and performance in every section of the network.
At a moe practical level, the fabric can also refer to the command and control of your network. You might have a network fabric software or system. In that case, you’re using the term “fabric” to refer to the software management platform you use to manage the network in general.
From a network design standpoint, the fabric is the layer of switches and routers that connects all of your spines in a data center. Looking at it this way, you can see exactly where the fabric fits in your topology and how changes to the fabric will impact the overall network.
To summarize all of these concepts, the point of a fabric is to help you understand the exact layout of your network structure and how you control or change that layout.
Types of Fabrics
With all of that said, there are a number of different types of fabrics, each applying the concept to more specific design ideas and netting benefits accordingly.
Theoretically, you could have an infinite number of different types of fabrics, but we can lock down the idea of varying fabrics by looking at three common examples: LAN, SAN, and IP.
LAN
A LAN fabric is one built on wired connections. To understand how fabrics work in this manner, we can consider a simple Ethernet network. Every device on the network is connected via an Ethernet cable.
Thus, the LAN fabric is the map of every device and Ethernet cable across the network. You could still think of the fabric as the software that lets you manage the network, but in most cases, you would say that the fabric is the overall connection topology.
In a LAN fabric, you’re using the network map to ensure that you have the level of interconnectivity that you desire. Typically, each switch is connected to every single other switch in the adjacent layer. It’s similar to spine-leaf architecture in this sense. In fact, it’s common to add a fabric layer at the top of a spine-leaf structure in a data center specifically to maximize interconnectivity and network resilience.
SAN
A SAN fabric takes the same idea but applies it more narrowly. In this case, you’re working with a storage area network — meaning the fabric is only applied to a single segment of the network. Specifically, this segment houses storage devices.
The idea of a storage fabric is to create robust interconnectivity within the SAN while still limiting access from devices outside the SAN. This helps to maintain network segmentation for security and data integrity purposes. Yet, devices inside the SAN benefit from greater network performance by taking advantage of the fabric concept.
IP
The last primary type of network fabric is an IP fabric. An IP fabric is similar to a LAN fabric, but it isn’t limited to wired connections. It can also include wireless meshes.
Overall, it’s still using the same idea of interconnectivity, but IP fabrics are capable of achieving high levels of network complexity. All the while, powerful fabric software management tools enable you to optimize these complicated networks to squeeze out every last metric of performance.
Additional Learning Center Resources